Classification of Proteus penneri lipopolysaccharides into core region serotypes

Proteus penneri, previously designated as P. vulgaris biogroup 1, was identified and named in 1982 by Hickman et al. [1] on the basis of low DNA relatedness to DNA of the biogroups 2 and 3 representatives and its phenotypic differences. Although these Gram-negative, peritrichously flagellated rods are less common among Proteus spp. clinical isolates than P. mirabilis strains (70–90 % of Proteus spp. infections) [2, 3], the frequency of their isolation from hospital patients keeps on growing [2, 4] and misidentification may further contribute to a lowered number of P. penneri isolation reports [3, 5]. The most common body sites of P. penneri strains isolation are wounds (of abdomen, foot, groin, hip and neck) and the urinary tract, especially of long-term catheterized patients or individuals with anatomical abnormalities within the tract [2, 4, 6, 7]. P. penneri strains were also isolated from: blood, fecal specimens, ankle ulcer, sacral decubitus, conjunctiva, subcutaneous thigh or cerebral abscess, skin lesion aspirate, abdominal drain fluid, diabetic foot ulcer, bronchoalveolar lavage fluid, a pulmonary artery catheter tip, cerebrospinal fluid, sputum and the center of struvite bladder stone [2‐4, 6, 7].

P. penneri produce many virulence factors which enable them to cause infections, e.g., urease, fimbriae and hemagglutinins, hemolysins, metalloproteases, flagella, siderophores and lipopolysaccharide (LPS) [2, 4]. LPS consists of three structurally different regions: lipid A (defined structurally only for one P. mirabilis mutant), core oligosaccharide (OS) and O-specific polysaccharide (OPS) [4, 8]. Until now, OPS has been the best structurally and serologically characterized region of P. penneri LPS, which also defines the serospecificity of smooth bacterial cells. Twenty-six different OPS structures have been identified for P. penneri strains so far, among which seven are common also to the other representatives of the genus [4, 9, 10]. The P. penneri core region is less structurally diverse than OPS but in contrast to other enterobacterial LPS core regions characterized by lager structural variability. Up to date, 12 different structures of the outer core region, accounting for the structural diversity of the P. penneri LPS core regions, were identified (Fig. 1) [4, 11]. The majority of tested P. penneri strains presented one major glycoform of the inner core region [11, 12] (Fig. 1). There are only two strains, P. penneri 12 and 42, which present glycoforms of the inner core region not identified in any other Proteus spp. LPSs [4, 11, 12]. Moreover, the heterogeneity of this LPS part may appear also within one strain, e.g., P. penneri 13 forms ten variants of its core-lipid A backbone [4]. The P. penneri classification scheme is based on the OPSs serospecificity. So far, P. penneri isolates have been classified into 17 Proteus O serogroups, among which 13 consist of these species representatives only [4, 9, 10, 13]. To have an insight into the serological specificity of both polysaccharide and oligosaccharide parts of P. penneri LPS, it is worth creating an additional scheme classifying P. penneri LPSs into serotypes of their core regions. A core types classification scheme which together with the O-types scheme may serve as a diagnostic tool facilitating the assignment of Proteus LPSs of clinical isolates into appropriate O and R serotypes. In the current work, the results of serological studies prove the existence of another five serotypes of P. penneri core regions, which is evidence of further structural variations within this part of Proteus spp. LPS.

In previous serological studies, sera specific to appropriate P. penneri strains were tested with a set of 40 P. penneri LPSs and the core region serospecificity for the majority of those antigens was determined [16, 17, 19, 20]. In the present work, the reactivities of the core regions of P. penneri LPSs 2, 11, 12, 13, 16, 17, 18, 19, 24, 26, 28, 31, 35, 36, 38, 60, 63, 75, 100, 103, 107, 112, 114, 115 and 124 with five sera specific to the strains P. penneri 17, 103 (with defined structures of LPS core regions [11]), 28, 60 and 124 were analyzed by ELISA and Western blot and classified into core serotypes. Each serum was adsorbed with a single cross-reacting antigen and tested again in PIH or in ELISA (P. penneri 17 antiserum) with all LPSs reacting with the serum to exclude further serological differences within the groups.

The serological results of the present study have allowed classifying 22 LPSs of P. penneri 2, 11, 12, 16, 17, 18, 19, 24, 28, 31, 35, 36, 38, 60, 63, 75, 100, 103, 107, 114, 115 and 124 into one of the five new core serotypes (Table 2, below the middle line). Four of the antigen groups contain LPSs with determined core region structures [11] (in Table 2 marked with*), and one contains only LPSs (P. penneri 60 and 63) with structurally unknown core regions. Table 2 includes all serotypes of LPS core regions formed on the basis of the results (Table 1; Figs. 2, 3) and previous serological studies [16, 17, 19, 20]. The studies were possible to perform owing to a unique set of anti-core sera obtained by: (1) the adsorption of serum against the bacterial whole cells by the LPS possessing the same O-polysaccharide as the strain homologous to the serum and different core region serotypes, (2) the immunization of rabbit with the conjugate of core oligosaccharide with diphtheria toxoid and (3) the immunization of rabbit with the whole cells of rough strains or of smooth strains but having the majority of LPS molecules unsubstituted with O-polysaccharide chains.

Assigning the P. penneri LPSs to the appropriate core serotype has not always been easy to perform. In some cases, one LPS cross-reacted with the appropriate serum similarly to homologous LPS in one method and weaker in another technique, e.g., P. penneri 112 LPS reacted with P. penneri 124 antiserum to the same titer as homologous LPS (Table 1) and in Western blot more weakly than homologous LPS (Fig. 2f). The differences in the antisera reactivity titers within one group of LPSs may also result from the different number of LPS molecules substituted and unsubstituted with OPS-chains, thus from the different access of antibodies to the core region.

Although P. penneri 28 LPS possesses the core region with an unknown structure, it was selected as a representative of serotype no. 7 (Table 2) because it was possible to obtain the serum specific to its core region. The reactions observed for this serum in Western blot (Fig. 2a) with P. penneri 28, 16, 18 and 31 LPSs were the most diverse in their intensity. The LPS exhibiting the most distinguished reaction was P. penneri 18 LPS, for which the ladder-like banding pattern was observed at the level corresponding to the higher-molecular LPS species. Smearing which appeared for P. penneri 28 and 16 LPSs at the level, typical for higher-molecular mass species, concerned unspecific reactions since both LPSs have no common fragments in the OPS parts [9]. What is more, P. vulgaris 55/57 OPS which is known to be structurally identical to the O-antigen of the P. penneri 28 LPS has not reacted with the tested serum (Fig. 2a). The method, which occurred to be very useful for proving that all tested antigens present a common core region serotype, was the serum adsorption. It allowed abolishing the reactions observed previously with the unadsorbed serum (Table 1; Fig. 2a). The procedure of antiserum adsorption, by LPS molecules coated on SRBC as well as on the bacterial cells, is a proven method for assessing serological similarities between antigens. Even though slight differences in reactivities have been observed for a group of LPSs with unadsorbed serum, the adsorption procedure usually appears to be helpful in explaining disputable results [16, 17].

The serum adsorption procedure also enabled classifying 11 LPS to the core serotype presented by P. penneri 17 LPS. It is the most numerous group among the analyzed LPSs, having one serotype of the core region, including five LPSs of the determined core region structures (P. penneri 2, 11, 17, 19 and 107). The structural studies showed that the core regions of the mentioned LPSs possess a unique type of linkage [(1S)-GaloNAc-(1 → 4,6)-α-GalN] occurring also in OSs of a few Proteus spp. LPSs, which had not previously been reported in natural glycopolymers [11]. This linkage probably occurs in OSs of the remaining P. penneri LPS presenting the same serotype as the P. penneri 17 core region.

LPSs P. penneri 60 and 63 present a core region serotype new for P. penneri strains tested so far [16, 17, 19, 20]. P. penneri 60 antiserum cross-reacted only with the low-molecular-mass species of P. penneri 63 (Fig. 2d). This result also stays in agreement with the structural data showing that in OPSs of both LPSs there is no common fragment except for the N-acetyl-fucosamine residue [-3)-α-l-FucpNAc-(1-] [9, 22]. It should be remembered that, in contrast to the anti-core sera, O-specific immunoglobulins predominate in the sera specific to the whole bacterial cells, including P. penneri 60 antiserum. It was reflected in ELISA by the higher titer of the serum with homologous LPS than that observed for P. penneri 63 LPS (Table 1). P. penneri 63 LPS was assigned to the core region serotype presented by P. penneri 60 LPS on the grounds of the Western blot results after the serum adsorption by P. penneri 63 LPS molecules (Fig. 3a). Abolishing the bands corresponding to the reactions with low-molecular-mass species of both tested LPSs and leaving the ladder-like banding pattern typical for the serum reaction with high-molecular-mass species of P. penneri 60 LPS (Fig. 3a, in a frame) clearly confirmed the suggestion that these LPSs present a similar serotype of the core region.

P. penneri 103 LPS was found to present the same core region serotype as P. penneri 75 LPS on the basis of the Western blot results obtained after the adsorption of P. penneri 103 antiserum by P. penneri 75 LPS (Fig. 3b). Only a slight reaction with slow-migrating bands of P. penneri 103 LPS was observed. Comparing the OPSs structures of both LPSs, it can be noticed that they differ only in the lateral substituents of the glucose residue and probably the lateral group (EtnP in P. penneri 103 OPS) was recognized by the immunoglobulins remaining in anti-P. penneri 103 serum after its adsorption by P. penneri 75 LPS (d-Glc residue except EtnP) (Fig. 3b, the reaction in a frame) [23].

The P. penneri 124 strain possesses LPS with a structurally unidentified core region, but its rough form decided about its selection for the studies. All serological data, the previous [16] and present ones (Table 1; Fig. 2f), indicate the P. penneri 124 LPS core region as serologically identical to the OS of P. penneri 12 LPS with a known core region structure (Fig. 4). Analyzing the results of P. penneri 124 antiserum and the previous results obtained for P. penneri 13 antiserum [16] in relation to the core region structures of P. penneri 12, 13 and 26 LPSs (Fig. 4) [11], it was confirmed that the common fragment of the core regions: α-GalN-(1 → 4)-α-GalA may be responsible for the observed cross-reactions. The results obtained for P. penneri 124 antiserum, after its adsorption by the tested P. penneri LPSs (data not showed), differ insignificantly from those obtained for P. penneri 13 antiserum [16]. Adsorption of P. penneri 13 antiserum with P. penneri 12 and 124 LPSs left a small fraction of antibodies (reciprocal titer 1.600) reacting with P. penneri 13, 112 and 26 LPSs, which was not observed in the present studies (all antibodies against P. penneri 12 and 124 LPSs were removed). It had been shown previously that those antibodies probably recognized the α-Hep-(1 → 2)-α-dd-Hep fragment, which is present in the core region of P. penneri 13 (probably of 112) and 26 LPSs but absent from P. penneri 12 LPS (probably from P. penneri 124) (Fig. 4). The weaker reaction of P. penneri 124 antiserum with P. penneri 26 LPS (Table 1; Fig. 2f) can be explained by: (1) the lack in the serum of the antibodies binding to the heptose disaccharide, (2) the additional terminal Glc residue (present only in the OS of P. penneri 26 LPS), absent from the outer core region of the remaining LPSs tested (Fig. 4), which may hamper binding the antibodies to the α-GalN-(1 → 4)-α-GalA fragment. However, this small structural difference in the P. penneri 13 and 26 core regions had not been shown to result in decreasing the titer of P. penneri 13 antiserum reactivity with P. penneri 26 LPS, compared to the titer in the homologous system [16]. Thus, the P. penneri 26 LPS was classified into the group of LPSs (with one core region serotype) represented by P. penneri 13 LPS as its subgroup 4a (Table 2).

Serological classification of Proteus spp. strains is not a typing method commonly used in clinical laboratories. However, the serological classification scheme based on the P. penneri core region provides useful information necessary for assigning the clinical isolates, not only from P. penneri species, to an appropriate core region serotype. Further extension of the scheme with other representatives may form a database, which, together with the scheme of Proteus LPS O-types, will show which core region serotype dominates within an area where the tested isolates come from. This may help in the selection of antigens presenting the most common O and R serotypes and serve as a tool to identify common epitopes among strains which may upon immunisation lead to the formation of cross-reactive and cross-protective antibodies against the core region as shown in one example for E. coli [24].